The core of the MEPER technology visual inspection intelligent system consists of three parts: image acquisition, image processing and analysis, and result output and control.

In the image acquisition phase, a high-resolution industrial camera paired with a professional optical lens swiftly captures the subtle features on the surface of hollow plastic bottles. Depending on the inspection requirements, different types of light source combinations are employed to provide optimal lighting conditions for the target object, highlighting the differences between defects and standard parts. For instance, when detecting minor scratches on the bottle body, high-angle ring light can create a unique reflection at the scratch site, facilitating clear capture by the camera.

In the image processing and analysis stage, AI algorithms are one of the highlights of this solution. Compared to traditional detection methods that are limited to fixed features and explicit indicators, the device, based on deep learning, can accurately identify various defects of the bottle body, such as black spots, stains, cracks, deformations on the surface, flatness and foreign objects on the sealing surface of the bottle cap, as well as issues such as dimensional deviations and thread integrity. By analyzing image data from different dimensions, the accuracy and reliability of detection results are ensured. In addition, product companies can establish a private database (supported by the cloud) through the device, which not only ensures the accurate and efficient deployment of AI models but also enables fine data management and control, ensuring data usage security.